A task has been running a long time
Wider scopes do substantially more work, and Entire system refreshes the whole organisation model. Check the scope you sent on.
If it is genuinely stuck rather than slow, the run's Activity shows where it stopped.
It produced far more than I wanted
Almost always scope. Re-run on Only this, which answers the request and caps how many artefacts it may produce.
It produced something thin and disconnected
The reverse. Only this deliberately does not pull in connected work — re-run on Connected.
It is waiting for an answer
It hit something it could not responsibly decide. Answer it and the run continues from where it paused; the answer is written into the work, not just recorded. See Pausing, resuming and answering.
A number looks wrong
Check Issues first: if it could not find a value it will have used a labelled placeholder and told you. Answering the data request replaces it and recalculates.
If the number is real but wrong, Activity shows where it came from.
It chose the wrong surface
Move the result to another surface — see Reviewing an interface. To be consulted before it builds, turn on Ask for permission at interface selection.
I have run out of credits
Work stops rather than degrading silently. See Usage and credits — and check whether you have been running wider scopes than you need.
The output does not sound like us
Set the directory's ontology so it uses your words, and check the brand book for anything visual.